摘要
给出了n阶隐马尔可夫模型(HMMn)的定义及结构。在传统的隐马尔可夫模型及二阶隐马尔可夫模型(HMM2)的基础上研究了HMMn的前向、后向算法,Baum-Welch算法,并导出了HMMn在单观测序列和多观测序列培训两种情况下的参数估计公式。
Definition and structure are given of nth-order hidden Markov models.Forward-backward algorithm and Baum-Welch algorithm of the models are studied based on the traditional second-order hidden Markov model.Parameter estimation equations for the models are derived for the cases of both single and multiple observation sequence training.
出处
《南京邮电大学学报(自然科学版)》
2011年第4期118-124,共7页
Journal of Nanjing University of Posts and Telecommunications:Natural Science Edition
基金
江苏省高校自然科学基金(08KJB510012)资助项目